Ye, Wenrui2025-01-202025-01-202025-01-202025-01-16https://hdl.handle.net/10012/21383The transition to renewable energy systems is critical in mitigating climate change and reducing fossil fuel dependence. However, integrating these variable and intermittent sources into the existing grid raises challenges such as dynamic energy demand management and resource underutilization, leading to increased operational costs and hindering broader adoption. This thesis develops algorithms to optimize renewable energy systems, enhancing their integration and operational efficiency. The research makes a significant contribution to enhancing the utilization, reliability, and economic viability of renewable energy systems, supporting a smoother transition to sustainable energy practices. This thesis first enhances the energy generation of photovoltaic panels by optimizing their tilt angles to maximize solar energy capture under varying environmental conditions. The machine learning models provided accurate predictions of photovoltaic output, allowing for data-driven insights into optimal system performance. A subsequent optimization process identified the best tilt angles during the operation. The results demonstrated an increase in annual energy output by up to 9.7% compared to fixed-tilt systems. This confirms that dynamic tilt adjustment is an effective strategy for maximizing photovoltaic energy generation. The second part of the research focuses on optimizing the capacity of renewable energy system components, with a particular emphasis on energy storage systems such as batteries. This project addressed the challenge of determining the optimal capacity for each component to efficiently meet energy demands while minimizing costs. A two-stage optimization approach was applied: first, a genetic algorithm generated candidate configurations with specific capacities; second, a simulation was conducted using an energy management algorithm to evaluate the performance of these configurations. The optimized configurations led to an overall system energy independence score of 0.51 and an 18.12% higher internal rate of return, validating the effectiveness of the integrated optimization approach in capacity planning and highlighting the importance of appropriately sized energy storage in enhancing system performance. The final part introduces an advanced energy management algorithm inspired by Model Predictive Control, integrating both batteries and hydrogen storage to enhance renewable energy utilization. By employing time series data and transformer-based models, the system accurately predicts future energy demand. These predictions enable a rolling window optimization technique that utilizes machine learning for dynamic energy management. The inclusion of hydrogen storage allows excess renewable energy to be stored as hydrogen, providing a versatile energy carrier for applications beyond electricity and improving overall renewable energy utilization. This approach improved demand forecasting accuracy by 41.21% and increased the adjusted green hydrogen production rate from 29.54% to 54.3%, demonstrating that advanced predictive energy management strategies, combined with diverse energy storage solutions, significantly enhance system adaptability, efficiency, and renewable energy utilization. These studies demonstrate that optimizing component configurations and energy management strategies—while integrating advanced energy storage systems like batteries and hydrogen—substantially improves the efficiency, reliability, and economic viability of renewable energy systems. The research provides valuable insights for integrating renewable energy into existing grids and supports the transition toward more sustainable and resilient energy infrastructures.enRenewable EnergyOptimizationSystem ModellingDoctoral Thesis